includes/Sanitizer.php

   1 <?php
   2 /**
   3  * XHTML sanitizer for MediaWiki
   4  *
   5  * Copyright (C) 2002-2005 Brion Vibber <brion@pobox.com> et al
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Parser
  25  */
  26
  27 /**
  28  * Regular expression to match various types of character references in
  29  * Sanitizer::normalizeCharReferences and Sanitizer::decodeCharReferences
  30  */
  31 define( 'MW_CHAR_REFS_REGEX',
  32         '/&([A-Za-z0-9\x80-\xff]+);
  33          |&\#([0-9]+);
  34          |&\#x([0-9A-Za-z]+);
  35          |&\#X([0-9A-Za-z]+);
  36          |(&)/x' );
  37
  38 /**
  39  * Regular expression to match HTML/XML attribute pairs within a tag.
  40  * Allows some... latitude.
  41  * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
  42  */
  43 $attrib = '[A-Za-z0-9]';
  44 $space = '[\x09\x0a\x0d\x20]';
  45 define( 'MW_ATTRIBS_REGEX',
  46         "/(?:^|$space)($attrib+)
  47           ($space*=$space*
  48                 (?:
  49                  # The attribute value: quoted or alone
  50                   \"([^<\"]*)\"
  51                  | '([^<']*)'
  52                  |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
  53                  |  (\#[0-9a-fA-F]+) # Technically wrong, but lots of
  54                                                          # colors are specified like this.
  55                                                          # We'll be normalizing it.
  56                 )
  57            )?(?=$space|\$)/sx" );
  58
  59 /**
  60  * List of all named character entities defined in HTML 4.01
  61  * http://www.w3.org/TR/html4/sgml/entities.html
  62  * This list does *not* include &apos;, which is part of XHTML
  63  * 1.0 but not HTML 4.01.  It is handled as a special case in
  64  * the code.
  65  * @private
  66  */
  67 global $wgHtmlEntities;
  68 $wgHtmlEntities = array(
  69         'Aacute'   => 193,
  70         'aacute'   => 225,
  71         'Acirc'    => 194,
  72         'acirc'    => 226,
  73         'acute'    => 180,
  74         'AElig'    => 198,
  75         'aelig'    => 230,
  76         'Agrave'   => 192,
  77         'agrave'   => 224,
  78         'alefsym'  => 8501,
  79         'Alpha'    => 913,
  80         'alpha'    => 945,
  81         'amp'      => 38,
  82         'and'      => 8743,
  83         'ang'      => 8736,
  84         'Aring'    => 197,
  85         'aring'    => 229,
  86         'asymp'    => 8776,
  87         'Atilde'   => 195,
  88         'atilde'   => 227,
  89         'Auml'     => 196,
  90         'auml'     => 228,
  91         'bdquo'    => 8222,
  92         'Beta'     => 914,
  93         'beta'     => 946,
  94         'brvbar'   => 166,
  95         'bull'     => 8226,
  96         'cap'      => 8745,
  97         'Ccedil'   => 199,
  98         'ccedil'   => 231,
  99         'cedil'    => 184,
 100         'cent'     => 162,
 101         'Chi'      => 935,
 102         'chi'      => 967,
 103         'circ'     => 710,
 104         'clubs'    => 9827,
 105         'cong'     => 8773,
 106         'copy'     => 169,
 107         'crarr'    => 8629,
 108         'cup'      => 8746,
 109         'curren'   => 164,
 110         'dagger'   => 8224,
 111         'Dagger'   => 8225,
 112         'darr'     => 8595,
 113         'dArr'     => 8659,
 114         'deg'      => 176,
 115         'Delta'    => 916,
 116         'delta'    => 948,
 117         'diams'    => 9830,
 118         'divide'   => 247,
 119         'Eacute'   => 201,
 120         'eacute'   => 233,
 121         'Ecirc'    => 202,
 122         'ecirc'    => 234,
 123         'Egrave'   => 200,
 124         'egrave'   => 232,
 125         'empty'    => 8709,
 126         'emsp'     => 8195,
 127         'ensp'     => 8194,
 128         'Epsilon'  => 917,
 129         'epsilon'  => 949,
 130         'equiv'    => 8801,
 131         'Eta'      => 919,
 132         'eta'      => 951,
 133         'ETH'      => 208,
 134         'eth'      => 240,
 135         'Euml'     => 203,
 136         'euml'     => 235,
 137         'euro'     => 8364,
 138         'exist'    => 8707,
 139         'fnof'     => 402,
 140         'forall'   => 8704,
 141         'frac12'   => 189,
 142         'frac14'   => 188,
 143         'frac34'   => 190,
 144         'frasl'    => 8260,
 145         'Gamma'    => 915,
 146         'gamma'    => 947,
 147         'ge'       => 8805,
 148         'gt'       => 62,
 149         'harr'     => 8596,
 150         'hArr'     => 8660,
 151         'hearts'   => 9829,
 152         'hellip'   => 8230,
 153         'Iacute'   => 205,
 154         'iacute'   => 237,
 155         'Icirc'    => 206,
 156         'icirc'    => 238,
 157         'iexcl'    => 161,
 158         'Igrave'   => 204,
 159         'igrave'   => 236,
 160         'image'    => 8465,
 161         'infin'    => 8734,
 162         'int'      => 8747,
 163         'Iota'     => 921,
 164         'iota'     => 953,
 165         'iquest'   => 191,
 166         'isin'     => 8712,
 167         'Iuml'     => 207,
 168         'iuml'     => 239,
 169         'Kappa'    => 922,
 170         'kappa'    => 954,
 171         'Lambda'   => 923,
 172         'lambda'   => 955,
 173         'lang'     => 9001,
 174         'laquo'    => 171,
 175         'larr'     => 8592,
 176         'lArr'     => 8656,
 177         'lceil'    => 8968,
 178         'ldquo'    => 8220,
 179         'le'       => 8804,
 180         'lfloor'   => 8970,
 181         'lowast'   => 8727,
 182         'loz'      => 9674,
 183         'lrm'      => 8206,
 184         'lsaquo'   => 8249,
 185         'lsquo'    => 8216,
 186         'lt'       => 60,
 187         'macr'     => 175,
 188         'mdash'    => 8212,
 189         'micro'    => 181,
 190         'middot'   => 183,
 191         'minus'    => 8722,
 192         'Mu'       => 924,
 193         'mu'       => 956,
 194         'nabla'    => 8711,
 195         'nbsp'     => 160,
 196         'ndash'    => 8211,
 197         'ne'       => 8800,
 198         'ni'       => 8715,
 199         'not'      => 172,
 200         'notin'    => 8713,
 201         'nsub'     => 8836,
 202         'Ntilde'   => 209,
 203         'ntilde'   => 241,
 204         'Nu'       => 925,
 205         'nu'       => 957,
 206         'Oacute'   => 211,
 207         'oacute'   => 243,
 208         'Ocirc'    => 212,
 209         'ocirc'    => 244,
 210         'OElig'    => 338,
 211         'oelig'    => 339,
 212         'Ograve'   => 210,
 213         'ograve'   => 242,
 214         'oline'    => 8254,
 215         'Omega'    => 937,
 216         'omega'    => 969,
 217         'Omicron'  => 927,
 218         'omicron'  => 959,
 219         'oplus'    => 8853,
 220         'or'       => 8744,
 221         'ordf'     => 170,
 222         'ordm'     => 186,
 223         'Oslash'   => 216,
 224         'oslash'   => 248,
 225         'Otilde'   => 213,
 226         'otilde'   => 245,
 227         'otimes'   => 8855,
 228         'Ouml'     => 214,
 229         'ouml'     => 246,
 230         'para'     => 182,
 231         'part'     => 8706,
 232         'permil'   => 8240,
 233         'perp'     => 8869,
 234         'Phi'      => 934,
 235         'phi'      => 966,
 236         'Pi'       => 928,
 237         'pi'       => 960,
 238         'piv'      => 982,
 239         'plusmn'   => 177,
 240         'pound'    => 163,
 241         'prime'    => 8242,
 242         'Prime'    => 8243,
 243         'prod'     => 8719,
 244         'prop'     => 8733,
 245         'Psi'      => 936,
 246         'psi'      => 968,
 247         'quot'     => 34,
 248         'radic'    => 8730,
 249         'rang'     => 9002,
 250         'raquo'    => 187,
 251         'rarr'     => 8594,
 252         'rArr'     => 8658,
 253         'rceil'    => 8969,
 254         'rdquo'    => 8221,
 255         'real'     => 8476,
 256         'reg'      => 174,
 257         'rfloor'   => 8971,
 258         'Rho'      => 929,
 259         'rho'      => 961,
 260         'rlm'      => 8207,
 261         'rsaquo'   => 8250,
 262         'rsquo'    => 8217,
 263         'sbquo'    => 8218,
 264         'Scaron'   => 352,
 265         'scaron'   => 353,
 266         'sdot'     => 8901,
 267         'sect'     => 167,
 268         'shy'      => 173,
 269         'Sigma'    => 931,
 270         'sigma'    => 963,
 271         'sigmaf'   => 962,
 272         'sim'      => 8764,
 273         'spades'   => 9824,
 274         'sub'      => 8834,
 275         'sube'     => 8838,
 276         'sum'      => 8721,
 277         'sup'      => 8835,
 278         'sup1'     => 185,
 279         'sup2'     => 178,
 280         'sup3'     => 179,
 281         'supe'     => 8839,
 282         'szlig'    => 223,
 283         'Tau'      => 932,
 284         'tau'      => 964,
 285         'there4'   => 8756,
 286         'Theta'    => 920,
 287         'theta'    => 952,
 288         'thetasym' => 977,
 289         'thinsp'   => 8201,
 290         'THORN'    => 222,
 291         'thorn'    => 254,
 292         'tilde'    => 732,
 293         'times'    => 215,
 294         'trade'    => 8482,
 295         'Uacute'   => 218,
 296         'uacute'   => 250,
 297         'uarr'     => 8593,
 298         'uArr'     => 8657,
 299         'Ucirc'    => 219,
 300         'ucirc'    => 251,
 301         'Ugrave'   => 217,
 302         'ugrave'   => 249,
 303         'uml'      => 168,
 304         'upsih'    => 978,
 305         'Upsilon'  => 933,
 306         'upsilon'  => 965,
 307         'Uuml'     => 220,
 308         'uuml'     => 252,
 309         'weierp'   => 8472,
 310         'Xi'       => 926,
 311         'xi'       => 958,
 312         'Yacute'   => 221,
 313         'yacute'   => 253,
 314         'yen'      => 165,
 315         'Yuml'     => 376,
 316         'yuml'     => 255,
 317         'Zeta'     => 918,
 318         'zeta'     => 950,
 319         'zwj'      => 8205,
 320         'zwnj'     => 8204 );
 321
 322 /**
 323  * Character entity aliases accepted by MediaWiki
 324  * XXX: decodeEntity() assumes that all values in this array are valid keys to $wgHtmlEntities
 325  */
 326 global $wgHtmlEntityAliases;
 327 $wgHtmlEntityAliases = array(
 328         'רלמ' => 'rlm',
 329         'رلم' => 'rlm',
 330 );
 331
 332
 333 /**
 334  * XHTML sanitizer for MediaWiki
 335  * @ingroup Parser
 336  */
 337 class Sanitizer {
 338         /**
 339          * Cleans up HTML, removes dangerous tags and attributes, and
 340          * removes HTML comments
 341          * @private
 342          * @param string $text
 343          * @param callback $processCallback to do any variable or parameter replacements in HTML attribute values
 344          * @param array $args for the processing callback
 345          * @return string
 346          */
 347         static function removeHTMLtags( $text, $processCallback = null, $args = array(), $extratags = array() ) {
 348                 global $wgUseTidy;
 349
 350                 static $htmlpairs, $htmlsingle, $htmlsingleonly, $htmlnest, $tabletags,
 351                         $htmllist, $listtags, $htmlsingleallowed, $htmlelements, $staticInitialised;
 352
 353                 wfProfileIn( __METHOD__ );
 354
 355                 if ( !$staticInitialised ) {
 356
 357                         $htmlpairs = array_merge( $extratags, array( # Tags that must be closed
 358                                 'b', 'del', 'i', 'ins', 'u', 'font', 'big', 'small', 'sub', 'sup', 'h1',
 359                                 'h2', 'h3', 'h4', 'h5', 'h6', 'cite', 'code', 'em', 's',
 360                                 'strike', 'strong', 'tt', 'var', 'div', 'center',
 361                                 'blockquote', 'ol', 'ul', 'dl', 'table', 'caption', 'pre',
 362                                 'ruby', 'rt' , 'rb' , 'rp', 'p', 'span', 'u'
 363                         ) );
 364                         $htmlsingle = array(
 365                                 'br', 'hr', 'li', 'dt', 'dd'
 366                         );
 367                         $htmlsingleonly = array( # Elements that cannot have close tags
 368                                 'br', 'hr'
 369                         );
 370                         $htmlnest = array( # Tags that can be nested--??
 371                                 'table', 'tr', 'td', 'th', 'div', 'blockquote', 'ol', 'ul',
 372                                 'dl', 'font', 'big', 'small', 'sub', 'sup', 'span'
 373                         );
 374                         $tabletags = array( # Can only appear inside table, we will close them
 375                                 'td', 'th', 'tr',
 376                         );
 377                         $htmllist = array( # Tags used by list
 378                                 'ul','ol',
 379                         );
 380                         $listtags = array( # Tags that can appear in a list
 381                                 'li',
 382                         );
 383
 384                         $htmlsingleallowed = array_merge( $htmlsingle, $tabletags );
 385                         $htmlelements = array_merge( $htmlsingle, $htmlpairs, $htmlnest );
 386
 387                         # Convert them all to hashtables for faster lookup
 388                         $vars = array( 'htmlpairs', 'htmlsingle', 'htmlsingleonly', 'htmlnest', 'tabletags',
 389                                 'htmllist', 'listtags', 'htmlsingleallowed', 'htmlelements' );
 390                         foreach ( $vars as $var ) {
 391                                 $$var = array_flip( $$var );
 392                         }
 393                         $staticInitialised = true;
 394                 }
 395
 396                 # Remove HTML comments
 397                 $text = Sanitizer::removeHTMLcomments( $text );
 398                 $bits = explode( '<', $text );
 399                 $text = str_replace( '>', '&gt;', array_shift( $bits ) );
 400                 if(!$wgUseTidy) {
 401                         $tagstack = $tablestack = array();
 402                         foreach ( $bits as $x ) {
 403                                 $regs = array();
 404                                 if( preg_match( '!^(/?)(\\w+)([^>]*?)(/{0,1}>)([^<]*)$!', $x, $regs ) ) {
 405                                         list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 406                                 } else {
 407                                         $slash = $t = $params = $brace = $rest = null;
 408                                 }
 409
 410                                 $badtag = 0 ;
 411                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 412                                         # Check our stack
 413                                         if ( $slash ) {
 414                                                 # Closing a tag...
 415                                                 if( isset( $htmlsingleonly[$t] ) ) {
 416                                                         $badtag = 1;
 417                                                 } elseif ( ( $ot = @array_pop( $tagstack ) ) != $t ) {
 418                                                         if ( isset( $htmlsingleallowed[$ot] ) ) {
 419                                                                 # Pop all elements with an optional close tag
 420                                                                 # and see if we find a match below them
 421                                                                 $optstack = array();
 422                                                                 array_push ($optstack, $ot);
 423                                                                 while ( ( ( $ot = @array_pop( $tagstack ) ) != $t ) &&
 424                                                                                 isset( $htmlsingleallowed[$ot] ) )
 425                                                                 {
 426                                                                         array_push ($optstack, $ot);
 427                                                                 }
 428                                                                 if ( $t != $ot ) {
 429                                                                         # No match. Push the optinal elements back again
 430                                                                         $badtag = 1;
 431                                                                         while ( $ot = @array_pop( $optstack ) ) {
 432                                                                                 array_push( $tagstack, $ot );
 433                                                                         }
 434                                                                 }
 435                                                         } else {
 436                                                                 @array_push( $tagstack, $ot );
 437                                                                 # <li> can be nested in <ul> or <ol>, skip those cases:
 438                                                                 if(!(isset( $htmllist[$ot] ) && isset( $listtags[$t] ) )) {
 439                                                                         $badtag = 1;
 440                                                                 }
 441                                                         }
 442                                                 } else {
 443                                                         if ( $t == 'table' ) {
 444                                                                 $tagstack = array_pop( $tablestack );
 445                                                         }
 446                                                 }
 447                                                 $newparams = '';
 448                                         } else {
 449                                                 # Keep track for later
 450                                                 if ( isset( $tabletags[$t] ) &&
 451                                                 ! in_array( 'table', $tagstack ) ) {
 452                                                         $badtag = 1;
 453                                                 } else if ( in_array( $t, $tagstack ) &&
 454                                                 ! isset( $htmlnest [$t ] ) ) {
 455                                                         $badtag = 1 ;
 456                                                 # Is it a self closed htmlpair ? (bug 5487)
 457                                                 } else if( $brace == '/>' &&
 458                                                 isset( $htmlpairs[$t] ) ) {
 459                                                         $badtag = 1;
 460                                                 } elseif( isset( $htmlsingleonly[$t] ) ) {
 461                                                         # Hack to force empty tag for uncloseable elements
 462                                                         $brace = '/>';
 463                                                 } else if( isset( $htmlsingle[$t] ) ) {
 464                                                         # Hack to not close $htmlsingle tags
 465                                                         $brace = NULL;
 466                                                 } else if( isset( $tabletags[$t] )
 467                                                 &&  in_array($t ,$tagstack) ) {
 468                                                         // New table tag but forgot to close the previous one
 469                                                         $text .= "</$t>";
 470                                                 } else {
 471                                                         if ( $t == 'table' ) {
 472                                                                 array_push( $tablestack, $tagstack );
 473                                                                 $tagstack = array();
 474                                                         }
 475                                                         array_push( $tagstack, $t );
 476                                                 }
 477
 478                                                 # Replace any variables or template parameters with
 479                                                 # plaintext results.
 480                                                 if( is_callable( $processCallback ) ) {
 481                                                         call_user_func_array( $processCallback, array( &$params, $args ) );
 482                                                 }
 483
 484                                                 # Strip non-approved attributes from the tag
 485                                                 $newparams = Sanitizer::fixTagAttributes( $params, $t );
 486                                         }
 487                                         if ( ! $badtag ) {
 488                                                 $rest = str_replace( '>', '&gt;', $rest );
 489                                                 $close = ( $brace == '/>' && !$slash ) ? ' /' : '';
 490                                                 $text .= "<$slash$t$newparams$close>$rest";
 491                                                 continue;
 492                                         }
 493                                 }
 494                                 $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 495                         }
 496                         # Close off any remaining tags
 497                         while ( is_array( $tagstack ) && ($t = array_pop( $tagstack )) ) {
 498                                 $text .= "</$t>\n";
 499                                 if ( $t == 'table' ) { $tagstack = array_pop( $tablestack ); }
 500                         }
 501                 } else {
 502                         # this might be possible using tidy itself
 503                         foreach ( $bits as $x ) {
 504                                 preg_match( '/^(\\/?)(\\w+)([^>]*?)(\\/{0,1}>)([^<]*)$/',
 505                                 $x, $regs );
 506                                 @list( /* $qbar */, $slash, $t, $params, $brace, $rest ) = $regs;
 507                                 if ( isset( $htmlelements[$t = strtolower( $t )] ) ) {
 508                                         if( is_callable( $processCallback ) ) {
 509                                                 call_user_func_array( $processCallback, array( &$params, $args ) );
 510                                         }
 511                                         $newparams = Sanitizer::fixTagAttributes( $params, $t );
 512                                         $rest = str_replace( '>', '&gt;', $rest );
 513                                         $text .= "<$slash$t$newparams$brace$rest";
 514                                 } else {
 515                                         $text .= '&lt;' . str_replace( '>', '&gt;', $x);
 516                                 }
 517                         }
 518                 }
 519                 wfProfileOut( __METHOD__ );
 520                 return $text;
 521         }
 522
 523         /**
 524          * Remove '<!--', '-->', and everything between.
 525          * To avoid leaving blank lines, when a comment is both preceded
 526          * and followed by a newline (ignoring spaces), trim leading and
 527          * trailing spaces and one of the newlines.
 528          *
 529          * @private
 530          * @param string $text
 531          * @return string
 532          */
 533         static function removeHTMLcomments( $text ) {
 534                 wfProfileIn( __METHOD__ );
 535                 while (($start = strpos($text, '<!--')) !== false) {
 536                         $end = strpos($text, '-->', $start + 4);
 537                         if ($end === false) {
 538                                 # Unterminated comment; bail out
 539                                 break;
 540                         }
 541
 542                         $end += 3;
 543
 544                         # Trim space and newline if the comment is both
 545                         # preceded and followed by a newline
 546                         $spaceStart = max($start - 1, 0);
 547                         $spaceLen = $end - $spaceStart;
 548                         while (substr($text, $spaceStart, 1) === ' ' && $spaceStart > 0) {
 549                                 $spaceStart--;
 550                                 $spaceLen++;
 551                         }
 552                         while (substr($text, $spaceStart + $spaceLen, 1) === ' ')
 553                                 $spaceLen++;
 554                         if (substr($text, $spaceStart, 1) === "\n" and substr($text, $spaceStart + $spaceLen, 1) === "\n") {
 555                                 # Remove the comment, leading and trailing
 556                                 # spaces, and leave only one newline.
 557                                 $text = substr_replace($text, "\n", $spaceStart, $spaceLen + 1);
 558                         }
 559                         else {
 560                                 # Remove just the comment.
 561                                 $text = substr_replace($text, '', $start, $end - $start);
 562                         }
 563                 }
 564                 wfProfileOut( __METHOD__ );
 565                 return $text;
 566         }
 567
 568         /**
 569          * Take an array of attribute names and values and normalize or discard
 570          * illegal values for the given element type.
 571          *
 572          * - Discards attributes not on a whitelist for the given element
 573          * - Unsafe style attributes are discarded
 574          * - Invalid id attributes are reencoded
 575          *
 576          * @param array $attribs
 577          * @param string $element
 578          * @return array
 579          *
 580          * @todo Check for legal values where the DTD limits things.
 581          * @todo Check for unique id attribute :P
 582          */
 583         static function validateTagAttributes( $attribs, $element ) {
 584                 return Sanitizer::validateAttributes( $attribs,
 585                         Sanitizer::attributeWhitelist( $element ) );
 586         }
 587
 588         /**
 589          * Take an array of attribute names and values and normalize or discard
 590          * illegal values for the given whitelist.
 591          *
 592          * - Discards attributes not the given whitelist
 593          * - Unsafe style attributes are discarded
 594          * - Invalid id attributes are reencoded
 595          *
 596          * @param array $attribs
 597          * @param array $whitelist list of allowed attribute names
 598          * @return array
 599          *
 600          * @todo Check for legal values where the DTD limits things.
 601          * @todo Check for unique id attribute :P
 602          */
 603         static function validateAttributes( $attribs, $whitelist ) {
 604                 $whitelist = array_flip( $whitelist );
 605                 $out = array();
 606                 foreach( $attribs as $attribute => $value ) {
 607                         if( !isset( $whitelist[$attribute] ) ) {
 608                                 continue;
 609                         }
 610                         # Strip javascript "expression" from stylesheets.
 611                         # http://msdn.microsoft.com/workshop/author/dhtml/overview/recalc.asp
 612                         if( $attribute == 'style' ) {
 613                                 $value = Sanitizer::checkCss( $value );
 614                                 if( $value === false ) {
 615                                         # haxx0r
 616                                         continue;
 617                                 }
 618                         }
 619
 620                         if ( $attribute === 'id' ) {
 621                                 global $wgEnforceHtmlIds;
 622                                 $value = Sanitizer::escapeId( $value,
 623                                         $wgEnforceHtmlIds ? array() : 'xml' );
 624                         }
 625
 626                         // If this attribute was previously set, override it.
 627                         // Output should only have one attribute of each name.
 628                         $out[$attribute] = $value;
 629                 }
 630                 return $out;
 631         }
 632
 633         /**
 634          * Merge two sets of HTML attributes.  Conflicting items in the second set
 635          * will override those in the first, except for 'class' attributes which
 636          * will be combined (if they're both strings).
 637          *
 638          * @todo implement merging for other attributes such as style
 639          * @param array $a
 640          * @param array $b
 641          * @return array
 642          */
 643         static function mergeAttributes( $a, $b ) {
 644                 $out = array_merge( $a, $b );
 645                 if( isset( $a['class'] ) && isset( $b['class'] )
 646                 && is_string( $a['class'] ) && is_string( $b['class'] )
 647                 && $a['class'] !== $b['class'] ) {
 648                         $classes = preg_split( '/\s+/', "{$a['class']} {$b['class']}",
 649                                 -1, PREG_SPLIT_NO_EMPTY );
 650                         $out['class'] = implode( ' ', array_unique( $classes ) );
 651                 }
 652                 return $out;
 653         }
 654
 655         /**
 656          * Pick apart some CSS and check it for forbidden or unsafe structures.
 657          * Returns a sanitized string, or false if it was just too evil.
 658          *
 659          * Currently URL references, 'expression', 'tps' are forbidden.
 660          *
 661          * @param string $value
 662          * @return mixed
 663          */
 664         static function checkCss( $value ) {
 665                 $stripped = Sanitizer::decodeCharReferences( $value );
 666
 667                 // Remove any comments; IE gets token splitting wrong
 668                 $stripped = StringUtils::delimiterReplace( '/*', '*/', ' ', $stripped );
 669
 670                 $value = $stripped;
 671
 672                 // ... and continue checks
 673                 $stripped = preg_replace( '!\\\\([0-9A-Fa-f]{1,6})[ \\n\\r\\t\\f]?!e',
 674                         'codepointToUtf8(hexdec("$1"))', $stripped );
 675                 $stripped = str_replace( '\\', '', $stripped );
 676                 if( preg_match( '/(?:expression|tps*:\/\/|url\\s*\().*/is',
 677                                 $stripped ) ) {
 678                         # haxx0r
 679                         return false;
 680                 }
 681
 682                 return $value;
 683         }
 684
 685         /**
 686          * Take a tag soup fragment listing an HTML element's attributes
 687          * and normalize it to well-formed XML, discarding unwanted attributes.
 688          * Output is safe for further wikitext processing, with escaping of
 689          * values that could trigger problems.
 690          *
 691          * - Normalizes attribute names to lowercase
 692          * - Discards attributes not on a whitelist for the given element
 693          * - Turns broken or invalid entities into plaintext
 694          * - Double-quotes all attribute values
 695          * - Attributes without values are given the name as attribute
 696          * - Double attributes are discarded
 697          * - Unsafe style attributes are discarded
 698          * - Prepends space if there are attributes.
 699          *
 700          * @param string $text
 701          * @param string $element
 702          * @return string
 703          */
 704         static function fixTagAttributes( $text, $element ) {
 705                 if( trim( $text ) == '' ) {
 706                         return '';
 707                 }
 708
 709                 $stripped = Sanitizer::validateTagAttributes(
 710                         Sanitizer::decodeTagAttributes( $text ), $element );
 711
 712                 $attribs = array();
 713                 foreach( $stripped as $attribute => $value ) {
 714                         $encAttribute = htmlspecialchars( $attribute );
 715                         $encValue = Sanitizer::safeEncodeAttribute( $value );
 716
 717                         $attribs[] = "$encAttribute=\"$encValue\"";
 718                 }
 719                 return count( $attribs ) ? ' ' . implode( ' ', $attribs ) : '';
 720         }
 721
 722         /**
 723          * Encode an attribute value for HTML output.
 724          * @param $text
 725          * @return HTML-encoded text fragment
 726          */
 727         static function encodeAttribute( $text ) {
 728                 $encValue = htmlspecialchars( $text, ENT_QUOTES );
 729
 730                 // Whitespace is normalized during attribute decoding,
 731                 // so if we've been passed non-spaces we must encode them
 732                 // ahead of time or they won't be preserved.
 733                 $encValue = strtr( $encValue, array(
 734                         "\n" => '&#10;',
 735                         "\r" => '&#13;',
 736                         "\t" => '&#9;',
 737                 ) );
 738
 739                 return $encValue;
 740         }
 741
 742         /**
 743          * Encode an attribute value for HTML tags, with extra armoring
 744          * against further wiki processing.
 745          * @param $text
 746          * @return HTML-encoded text fragment
 747          */
 748         static function safeEncodeAttribute( $text ) {
 749                 $encValue = Sanitizer::encodeAttribute( $text );
 750
 751                 # Templates and links may be expanded in later parsing,
 752                 # creating invalid or dangerous output. Suppress this.
 753                 $encValue = strtr( $encValue, array(
 754                         '<'    => '&lt;',   // This should never happen,
 755                         '>'    => '&gt;',   // we've received invalid input
 756                         '"'    => '&quot;', // which should have been escaped.
 757                         '{'    => '&#123;',
 758                         '['    => '&#91;',
 759                         "''"   => '&#39;&#39;',
 760                         'ISBN' => '&#73;SBN',
 761                         'RFC'  => '&#82;FC',
 762                         'PMID' => '&#80;MID',
 763                         '|'    => '&#124;',
 764                         '__'   => '&#95;_',
 765                 ) );
 766
 767                 # Stupid hack
 768                 $encValue = preg_replace_callback(
 769                         '/(' . wfUrlProtocols() . ')/',
 770                         array( 'Sanitizer', 'armorLinksCallback' ),
 771                         $encValue );
 772                 return $encValue;
 773         }
 774
 775         /**
 776          * Given a value escape it so that it can be used in an id attribute and
 777          * return it, this does not validate the value however (see first link)
 778          *
 779          * @see http://www.w3.org/TR/html401/types.html#type-name Valid characters
 780          *                                                          in the id and
 781          *                                                          name attributes
 782          * @see http://www.w3.org/TR/html401/struct/links.html#h-12.2.3 Anchors with the id attribute
 783          *
 784          * @param string $id      Id to validate
 785          * @param mixed  $options String or array of strings (default is array()):
 786          *   'noninitial': This is a non-initial fragment of an id, not a full id,
 787          *       so don't pay attention if the first character isn't valid at the
 788          *       beginning of an id.
 789          *   'xml': Don't restrict the id to be HTML4-compatible.  This option
 790          *       allows any alphabetic character to be used, per the XML standard.
 791          *       Therefore, it also completely changes the type of escaping: instead
 792          *       of weird dot-encoding, runs of invalid characters (mostly
 793          *       whitespace) are just compressed into a single underscore.
 794          * @return string
 795          */
 796         static function escapeId( $id, $options = array() ) {
 797                 $options = (array)$options;
 798
 799                 if ( !in_array( 'xml', $options ) ) {
 800                         # HTML4-style escaping
 801                         static $replace = array(
 802                                 '%3A' => ':',
 803                                 '%' => '.'
 804                         );
 805
 806                         $id = urlencode( Sanitizer::decodeCharReferences( strtr( $id, ' ', '_' ) ) );
 807                         $id = str_replace( array_keys( $replace ), array_values( $replace ), $id );
 808
 809                         /*
 810                         // Let's not go changing our links just yet, since we plan to
 811                         // rip this out anyway and switch to the prettier Unicode-friendly
 812                         // links as XHTML and HTML 5 allow and browsers seem to think is
 813                         // just a-ok to do.
 814                         //
 815                         if ( !preg_match( '/^[a-zA-Z]/', $id )
 816                         && !in_array( 'noninitial', $options ) )  {
 817                                 // Initial character must be a letter!
 818                                 $id = "x$id";
 819                         }
 820                         */
 821                         return $id;
 822                 }
 823
 824                 # XML-style escaping.  For the patterns used, see the XML 1.0 standard,
 825                 # 5th edition, NameStartChar and NameChar: <http://www.w3.org/TR/REC-xml/>
 826                 $nameStartChar = ':a-zA-Z_\xC0-\xD6\xD8-\xF6\xF8-\x{2FF}\x{370}-\x{37D}'
 827                         . '\x{37F}-\x{1FFF}\x{200C}-\x{200D}\x{2070}-\x{218F}\x{2C00}-\x{2FEF}'
 828                         . '\x{3001}-\x{D7FF}\x{F900}-\x{FDCF}\x{FDF0}-\x{FFFD}\x{10000}-\x{EFFFF}';
 829                 $nameChar = $nameStartChar . '.\-0-9\xB7\x{0300}-\x{036F}'
 830                         . '\x{203F}-\x{2040}';
 831                 # Replace _ as well so we don't get multiple consecutive underscores
 832                 $id = preg_replace( "/([^$nameChar]|_)+/u", '_', $id );
 833                 $id = trim( $id, '_' );
 834
 835                 if ( !preg_match( "/^[$nameStartChar]/u", $id )
 836                 && !in_array( 'noninitial', $options ) ) {
 837                         $id = "_$id";
 838                 }
 839
 840                 return $id;
 841         }
 842
 843         /**
 844          * Given a value, escape it so that it can be used as a CSS class and
 845          * return it.
 846          *
 847          * @todo For extra validity, input should be validated UTF-8.
 848          *
 849          * @see http://www.w3.org/TR/CSS21/syndata.html Valid characters/format
 850          *
 851          * @param string $class
 852          * @return string
 853          */
 854         static function escapeClass( $class ) {
 855                 // Convert ugly stuff to underscores and kill underscores in ugly places
 856                 return rtrim(preg_replace(
 857                         array('/(^[0-9\\-])|[\\x00-\\x20!"#$%&\'()*+,.\\/:;<=>?@[\\]^`{|}~]|\\xC2\\xA0/','/_+/'),
 858                         '_',
 859                         $class ), '_');
 860         }
 861
 862         /**
 863          * Given HTML input, escape with htmlspecialchars but un-escape entites.
 864          * This allows (generally harmless) entities like &nbsp; to survive.
 865          *
 866          * @param  string $html String to escape
 867          * @return string Escaped input
 868          */
 869         static function escapeHtmlAllowEntities( $html ) {
 870                 # It seems wise to escape ' as well as ", as a matter of course.  Can't
 871                 # hurt.
 872                 $html = htmlspecialchars( $html, ENT_QUOTES );
 873                 $html = str_replace( '&amp;', '&', $html );
 874                 $html = Sanitizer::normalizeCharReferences( $html );
 875                 return $html;
 876         }
 877
 878         /**
 879          * Regex replace callback for armoring links against further processing.
 880          * @param array $matches
 881          * @return string
 882          * @private
 883          */
 884         private static function armorLinksCallback( $matches ) {
 885                 return str_replace( ':', '&#58;', $matches[1] );
 886         }
 887
 888         /**
 889          * Return an associative array of attribute names and values from
 890          * a partial tag string. Attribute names are forces to lowercase,
 891          * character references are decoded to UTF-8 text.
 892          *
 893          * @param string
 894          * @return array
 895          */
 896         public static function decodeTagAttributes( $text ) {
 897                 $attribs = array();
 898
 899                 if( trim( $text ) == '' ) {
 900                         return $attribs;
 901                 }
 902
 903                 $pairs = array();
 904                 if( !preg_match_all(
 905                         MW_ATTRIBS_REGEX,
 906                         $text,
 907                         $pairs,
 908                         PREG_SET_ORDER ) ) {
 909                         return $attribs;
 910                 }
 911
 912                 foreach( $pairs as $set ) {
 913                         $attribute = strtolower( $set[1] );
 914                         $value = Sanitizer::getTagAttributeCallback( $set );
 915
 916                         // Normalize whitespace
 917                         $value = preg_replace( '/[\t\r\n ]+/', ' ', $value );
 918                         $value = trim( $value );
 919
 920                         // Decode character references
 921                         $attribs[$attribute] = Sanitizer::decodeCharReferences( $value );
 922                 }
 923                 return $attribs;
 924         }
 925
 926         /**
 927          * Pick the appropriate attribute value from a match set from the
 928          * MW_ATTRIBS_REGEX matches.
 929          *
 930          * @param array $set
 931          * @return string
 932          * @private
 933          */
 934         private static function getTagAttributeCallback( $set ) {
 935                 if( isset( $set[6] ) ) {
 936                         # Illegal #XXXXXX color with no quotes.
 937                         return $set[6];
 938                 } elseif( isset( $set[5] ) ) {
 939                         # No quotes.
 940                         return $set[5];
 941                 } elseif( isset( $set[4] ) ) {
 942                         # Single-quoted
 943                         return $set[4];
 944                 } elseif( isset( $set[3] ) ) {
 945                         # Double-quoted
 946                         return $set[3];
 947                 } elseif( !isset( $set[2] ) ) {
 948                         # In XHTML, attributes must have a value.
 949                         # For 'reduced' form, return explicitly the attribute name here.
 950                         return $set[1];
 951                 } else {
 952                         throw new MWException( "Tag conditions not met. This should never happen and is a bug." );
 953                 }
 954         }
 955
 956         /**
 957          * Normalize whitespace and character references in an XML source-
 958          * encoded text for an attribute value.
 959          *
 960          * See http://www.w3.org/TR/REC-xml/#AVNormalize for background,
 961          * but note that we are not returning the value, but are returning
 962          * XML source fragments that will be slapped into output.
 963          *
 964          * @param string $text
 965          * @return string
 966          * @private
 967          */
 968         private static function normalizeAttributeValue( $text ) {
 969                 return str_replace( '"', '&quot;',
 970                         self::normalizeWhitespace(
 971                                 Sanitizer::normalizeCharReferences( $text ) ) );
 972         }
 973
 974         private static function normalizeWhitespace( $text ) {
 975                 return preg_replace(
 976                         '/\r\n|[\x20\x0d\x0a\x09]/',
 977                         ' ',
 978                         $text );
 979         }
 980
 981         /**
 982          * Ensure that any entities and character references are legal
 983          * for XML and XHTML specifically. Any stray bits will be
 984          * &amp;-escaped to result in a valid text fragment.
 985          *
 986          * a. any named char refs must be known in XHTML
 987          * b. any numeric char refs must be legal chars, not invalid or forbidden
 988          * c. use &#x, not &#X
 989          * d. fix or reject non-valid attributes
 990          *
 991          * @param string $text
 992          * @return string
 993          * @private
 994          */
 995         static function normalizeCharReferences( $text ) {
 996                 return preg_replace_callback(
 997                         MW_CHAR_REFS_REGEX,
 998                         array( 'Sanitizer', 'normalizeCharReferencesCallback' ),
 999                         $text );
1000         }
1001         /**
1002          * @param string $matches
1003          * @return string
1004          */
1005         static function normalizeCharReferencesCallback( $matches ) {
1006                 $ret = null;
1007                 if( $matches[1] != '' ) {
1008                         $ret = Sanitizer::normalizeEntity( $matches[1] );
1009                 } elseif( $matches[2] != '' ) {
1010                         $ret = Sanitizer::decCharReference( $matches[2] );
1011                 } elseif( $matches[3] != ''  ) {
1012                         $ret = Sanitizer::hexCharReference( $matches[3] );
1013                 } elseif( $matches[4] != '' ) {
1014                         $ret = Sanitizer::hexCharReference( $matches[4] );
1015                 }
1016                 if( is_null( $ret ) ) {
1017                         return htmlspecialchars( $matches[0] );
1018                 } else {
1019                         return $ret;
1020                 }
1021         }
1022
1023         /**
1024          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1025          * return the named entity reference as is. If the entity is a
1026          * MediaWiki-specific alias, returns the HTML equivalent. Otherwise,
1027          * returns HTML-escaped text of pseudo-entity source (eg &amp;foo;)
1028          *
1029          * @param string $name
1030          * @return string
1031          * @static
1032          */
1033         static function normalizeEntity( $name ) {
1034                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1035                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1036                         return "&{$wgHtmlEntityAliases[$name]};";
1037                 } elseif( isset( $wgHtmlEntities[$name] ) ) {
1038                         return "&$name;";
1039                 } elseif( $name == 'apos' ) {
1040                         return "&#39;";  // "&apos;" is valid in XHTML, but not in HTML4
1041                 } else {
1042                         return "&amp;$name;";
1043                 }
1044         }
1045
1046         static function decCharReference( $codepoint ) {
1047                 $point = intval( $codepoint );
1048                 if( Sanitizer::validateCodepoint( $point ) ) {
1049                         return sprintf( '&#%d;', $point );
1050                 } else {
1051                         return null;
1052                 }
1053         }
1054
1055         static function hexCharReference( $codepoint ) {
1056                 $point = hexdec( $codepoint );
1057                 if( Sanitizer::validateCodepoint( $point ) ) {
1058                         return sprintf( '&#x%x;', $point );
1059                 } else {
1060                         return null;
1061                 }
1062         }
1063
1064         /**
1065          * Returns true if a given Unicode codepoint is a valid character in XML.
1066          * @param int $codepoint
1067          * @return bool
1068          */
1069         private static function validateCodepoint( $codepoint ) {
1070                 return ($codepoint ==    0x09)
1071                         || ($codepoint ==    0x0a)
1072                         || ($codepoint ==    0x0d)
1073                         || ($codepoint >=    0x20 && $codepoint <=   0xd7ff)
1074                         || ($codepoint >=  0xe000 && $codepoint <=   0xfffd)
1075                         || ($codepoint >= 0x10000 && $codepoint <= 0x10ffff);
1076         }
1077
1078         /**
1079          * Decode any character references, numeric or named entities,
1080          * in the text and return a UTF-8 string.
1081          *
1082          * @param string $text
1083          * @return string
1084          * @public
1085          * @static
1086          */
1087         public static function decodeCharReferences( $text ) {
1088                 return preg_replace_callback(
1089                         MW_CHAR_REFS_REGEX,
1090                         array( 'Sanitizer', 'decodeCharReferencesCallback' ),
1091                         $text );
1092         }
1093
1094         /**
1095          * @param string $matches
1096          * @return string
1097          */
1098         static function decodeCharReferencesCallback( $matches ) {
1099                 if( $matches[1] != '' ) {
1100                         return Sanitizer::decodeEntity( $matches[1] );
1101                 } elseif( $matches[2] != '' ) {
1102                         return  Sanitizer::decodeChar( intval( $matches[2] ) );
1103                 } elseif( $matches[3] != ''  ) {
1104                         return  Sanitizer::decodeChar( hexdec( $matches[3] ) );
1105                 } elseif( $matches[4] != '' ) {
1106                         return  Sanitizer::decodeChar( hexdec( $matches[4] ) );
1107                 }
1108                 # Last case should be an ampersand by itself
1109                 return $matches[0];
1110         }
1111
1112         /**
1113          * Return UTF-8 string for a codepoint if that is a valid
1114          * character reference, otherwise U+FFFD REPLACEMENT CHARACTER.
1115          * @param int $codepoint
1116          * @return string
1117          * @private
1118          */
1119         static function decodeChar( $codepoint ) {
1120                 if( Sanitizer::validateCodepoint( $codepoint ) ) {
1121                         return codepointToUtf8( $codepoint );
1122                 } else {
1123                         return UTF8_REPLACEMENT;
1124                 }
1125         }
1126
1127         /**
1128          * If the named entity is defined in the HTML 4.0/XHTML 1.0 DTD,
1129          * return the UTF-8 encoding of that character. Otherwise, returns
1130          * pseudo-entity source (eg &foo;)
1131          *
1132          * @param string $name
1133          * @return string
1134          */
1135         static function decodeEntity( $name ) {
1136                 global $wgHtmlEntities, $wgHtmlEntityAliases;
1137                 if ( isset( $wgHtmlEntityAliases[$name] ) ) {
1138                         $name = $wgHtmlEntityAliases[$name];
1139                 }
1140                 if( isset( $wgHtmlEntities[$name] ) ) {
1141                         return codepointToUtf8( $wgHtmlEntities[$name] );
1142                 } elseif( $name == 'apos' ) {
1143                         return "'";  // "&apos;" is not in $wgHtmlEntities, but it's still valid XHTML
1144                 } else {
1145                         return "&$name;";
1146                 }
1147         }
1148
1149         /**
1150          * Fetch the whitelist of acceptable attributes for a given
1151          * element name.
1152          *
1153          * @param string $element
1154          * @return array
1155          */
1156         static function attributeWhitelist( $element ) {
1157                 static $list;
1158                 if( !isset( $list ) ) {
1159                         $list = Sanitizer::setupAttributeWhitelist();
1160                 }
1161                 return isset( $list[$element] )
1162                         ? $list[$element]
1163                         : array();
1164         }
1165
1166         /**
1167          * Foreach array key (an allowed HTML element), return an array
1168          * of allowed attributes
1169          * @return array
1170          */
1171         static function setupAttributeWhitelist() {
1172                 $common = array( 'id', 'class', 'lang', 'dir', 'title', 'style' );
1173                 $block = array_merge( $common, array( 'align' ) );
1174                 $tablealign = array( 'align', 'char', 'charoff', 'valign' );
1175                 $tablecell = array( 'abbr',
1176                                     'axis',
1177                                     'headers',
1178                                     'scope',
1179                                     'rowspan',
1180                                     'colspan',
1181                                     'nowrap', # deprecated
1182                                     'width',  # deprecated
1183                                     'height', # deprecated
1184                                     'bgcolor' # deprecated
1185                                     );
1186
1187                 # Numbers refer to sections in HTML 4.01 standard describing the element.
1188                 # See: http://www.w3.org/TR/html4/
1189                 $whitelist = array (
1190                         # 7.5.4
1191                         'div'        => $block,
1192                         'center'     => $common, # deprecated
1193                         'span'       => $block, # ??
1194
1195                         # 7.5.5
1196                         'h1'         => $block,
1197                         'h2'         => $block,
1198                         'h3'         => $block,
1199                         'h4'         => $block,
1200                         'h5'         => $block,
1201                         'h6'         => $block,
1202
1203                         # 7.5.6
1204                         # address
1205
1206                         # 8.2.4
1207                         # bdo
1208
1209                         # 9.2.1
1210                         'em'         => $common,
1211                         'strong'     => $common,
1212                         'cite'       => $common,
1213                         # dfn
1214                         'code'       => $common,
1215                         # samp
1216                         # kbd
1217                         'var'        => $common,
1218                         # abbr
1219                         # acronym
1220
1221                         # 9.2.2
1222                         'blockquote' => array_merge( $common, array( 'cite' ) ),
1223                         # q
1224
1225                         # 9.2.3
1226                         'sub'        => $common,
1227                         'sup'        => $common,
1228
1229                         # 9.3.1
1230                         'p'          => $block,
1231
1232                         # 9.3.2
1233                         'br'         => array( 'id', 'class', 'title', 'style', 'clear' ),
1234
1235                         # 9.3.4
1236                         'pre'        => array_merge( $common, array( 'width' ) ),
1237
1238                         # 9.4
1239                         'ins'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1240                         'del'        => array_merge( $common, array( 'cite', 'datetime' ) ),
1241
1242                         # 10.2
1243                         'ul'         => array_merge( $common, array( 'type' ) ),
1244                         'ol'         => array_merge( $common, array( 'type', 'start' ) ),
1245                         'li'         => array_merge( $common, array( 'type', 'value' ) ),
1246
1247                         # 10.3
1248                         'dl'         => $common,
1249                         'dd'         => $common,
1250                         'dt'         => $common,
1251
1252                         # 11.2.1
1253                         'table'      => array_merge( $common,
1254                                                                 array( 'summary', 'width', 'border', 'frame',
1255                                                                                 'rules', 'cellspacing', 'cellpadding',
1256                                                                                 'align', 'bgcolor',
1257                                                                 ) ),
1258
1259                         # 11.2.2
1260                         'caption'    => array_merge( $common, array( 'align' ) ),
1261
1262                         # 11.2.3
1263                         'thead'      => array_merge( $common, $tablealign ),
1264                         'tfoot'      => array_merge( $common, $tablealign ),
1265                         'tbody'      => array_merge( $common, $tablealign ),
1266
1267                         # 11.2.4
1268                         'colgroup'   => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1269                         'col'        => array_merge( $common, array( 'span', 'width' ), $tablealign ),
1270
1271                         # 11.2.5
1272                         'tr'         => array_merge( $common, array( 'bgcolor' ), $tablealign ),
1273
1274                         # 11.2.6
1275                         'td'         => array_merge( $common, $tablecell, $tablealign ),
1276                         'th'         => array_merge( $common, $tablecell, $tablealign ),
1277
1278                         # 13.2
1279                         # Not usually allowed, but may be used for extension-style hooks
1280                         # such as <math> when it is rasterized
1281                         'img'        => array_merge( $common, array( 'alt' ) ),
1282
1283                         # 15.2.1
1284                         'tt'         => $common,
1285                         'b'          => $common,
1286                         'i'          => $common,
1287                         'big'        => $common,
1288                         'small'      => $common,
1289                         'strike'     => $common,
1290                         's'          => $common,
1291                         'u'          => $common,
1292
1293                         # 15.2.2
1294                         'font'       => array_merge( $common, array( 'size', 'color', 'face' ) ),
1295                         # basefont
1296
1297                         # 15.3
1298                         'hr'         => array_merge( $common, array( 'noshade', 'size', 'width' ) ),
1299
1300                         # XHTML Ruby annotation text module, simple ruby only.
1301                         # http://www.w3c.org/TR/ruby/
1302                         'ruby'       => $common,
1303                         # rbc
1304                         # rtc
1305                         'rb'         => $common,
1306                         'rt'         => $common, #array_merge( $common, array( 'rbspan' ) ),
1307                         'rp'         => $common,
1308
1309                         # MathML root element, where used for extensions
1310                         # 'title' may not be 100% valid here; it's XHTML
1311                         # http://www.w3.org/TR/REC-MathML/
1312                         'math'       => array( 'class', 'style', 'id', 'title' ),
1313                         );
1314                 return $whitelist;
1315         }
1316
1317         /**
1318          * Take a fragment of (potentially invalid) HTML and return
1319          * a version with any tags removed, encoded as plain text.
1320          *
1321          * Warning: this return value must be further escaped for literal
1322          * inclusion in HTML output as of 1.10!
1323          *
1324          * @param string $text HTML fragment
1325          * @return string
1326          */
1327         static function stripAllTags( $text ) {
1328                 # Actual <tags>
1329                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
1330
1331                 # Normalize &entities and whitespace
1332                 $text = self::decodeCharReferences( $text );
1333                 $text = self::normalizeWhitespace( $text );
1334
1335                 return $text;
1336         }
1337
1338         /**
1339          * Hack up a private DOCTYPE with HTML's standard entity declarations.
1340          * PHP 4 seemed to know these if you gave it an HTML doctype, but
1341          * PHP 5.1 doesn't.
1342          *
1343          * Use for passing XHTML fragments to PHP's XML parsing functions
1344          *
1345          * @return string
1346          * @static
1347          */
1348         static function hackDocType() {
1349                 global $wgHtmlEntities;
1350                 $out = "<!DOCTYPE html [\n";
1351                 foreach( $wgHtmlEntities as $entity => $codepoint ) {
1352                         $out .= "<!ENTITY $entity \"&#$codepoint;\">";
1353                 }
1354                 $out .= "]>\n";
1355                 return $out;
1356         }
1357
1358         static function cleanUrl( $url ) {
1359                 # Normalize any HTML entities in input. They will be
1360                 # re-escaped by makeExternalLink().
1361                 $url = Sanitizer::decodeCharReferences( $url );
1362
1363                 # Escape any control characters introduced by the above step
1364                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1365
1366                 # Validate hostname portion
1367                 $matches = array();
1368                 if( preg_match( '!^([^:]+:)(//[^/]+)?(.*)$!iD', $url, $matches ) ) {
1369                         list( /* $whole */, $protocol, $host, $rest ) = $matches;
1370
1371                         // Characters that will be ignored in IDNs.
1372                         // http://tools.ietf.org/html/3454#section-3.1
1373                         // Strip them before further processing so blacklists and such work.
1374                         $strip = "/
1375                                 \\s|          # general whitespace
1376                                 \xc2\xad|     # 00ad SOFT HYPHEN
1377                                 \xe1\xa0\x86| # 1806 MONGOLIAN TODO SOFT HYPHEN
1378                                 \xe2\x80\x8b| # 200b ZERO WIDTH SPACE
1379                                 \xe2\x81\xa0| # 2060 WORD JOINER
1380                                 \xef\xbb\xbf| # feff ZERO WIDTH NO-BREAK SPACE
1381                                 \xcd\x8f|     # 034f COMBINING GRAPHEME JOINER
1382                                 \xe1\xa0\x8b| # 180b MONGOLIAN FREE VARIATION SELECTOR ONE
1383                                 \xe1\xa0\x8c| # 180c MONGOLIAN FREE VARIATION SELECTOR TWO
1384                                 \xe1\xa0\x8d| # 180d MONGOLIAN FREE VARIATION SELECTOR THREE
1385                                 \xe2\x80\x8c| # 200c ZERO WIDTH NON-JOINER
1386                                 \xe2\x80\x8d| # 200d ZERO WIDTH JOINER
1387                                 [\xef\xb8\x80-\xef\xb8\x8f] # fe00-fe00f VARIATION SELECTOR-1-16
1388                                 /xuD";
1389
1390                         $host = preg_replace( $strip, '', $host );
1391
1392                         // @fixme: validate hostnames here
1393
1394                         return $protocol . $host . $rest;
1395                 } else {
1396                         return $url;
1397                 }
1398         }
1399
1400 }